A method and system for intelligent monitoring and dynamic resilience evaluation of a whole process of a cross-strait passenger and roll transportation

Abstract

The present application belongs to the technical field of passenger-rolling transportation, and particularly relates to a method and system for intelligent monitoring and dynamic resilience evaluation of cross-strait passenger-rolling transportation whole process. The method comprises: collecting multi-source data of cross-strait passenger-rolling transportation whole process and performing intelligent preprocessing, and constructing a five-dimensional dynamic correlation database; based on the five-dimensional dynamic correlation database and facility capacity coupling optimization evaluation model, combined with NSGA-III algorithm and DQN algorithm, obtaining facility capacity evaluation results of whole link peak period; based on the facility capacity evaluation results of whole link peak period and the five-dimensional dynamic correlation database, through a hybrid root cause model, obtaining short board positioning and root cause tracing results; based on the short board positioning and root cause tracing results and a seven-dimensional dynamic resilience evaluation model, using a dynamic weight distribution mechanism, obtaining resilience evaluation results; constructing an intelligent agent driven whole link autonomous iterative closed loop process, based on the resilience evaluation results, realizing autonomous generation, execution, feedback and iteration of peak period optimization scheme.

Description

Technical Field

[0001] This invention belongs to the field of passenger and roll-on / roll-off (Ro-Ro) transportation technology, specifically relating to a method and system for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off transportation. Background Technology

[0002] Cross-strait passenger and cargo ferry services serve as a core hub connecting the flow of people and goods across the Taiwan Strait. The transportation process encompasses multiple closely interconnected links, including port-backed collection and distribution channels, hinterland collection and distribution channels, vehicle waiting areas, terminal scheduling, passenger and cargo ferry transport, and transport from the opposite shore terminal. The transportation structure includes diverse types such as passengers, cars, and trucks (including vehicles transporting hazardous materials), and exhibits significant peak-valley flow characteristics. Peak traffic during holidays can reach 2-3 times the daily average, and in extreme cases, sudden demand may exceed 3 times the daily average. Simultaneously, cross-strait transportation faces multiple uncertainties, including complex waterway navigation environments (intersecting navigation, high density), natural disturbances (heavy fog, strong winds, typhoons, etc.), equipment failures, and policy adjustments. Coupled with the surge in traffic during peak periods, this places extremely high demands on the stable operation, emergency response, continuous optimization, and green and low-carbon development capabilities of the transportation system.

[0003] Currently, there are many shortcomings in the assessment technology related to cross-strait passenger and vehicle roll-on / roll-off (Ro-Ro) transportation, and there is a lack of specialized assessment capabilities for peak-season transportation. The core problems are as follows: 1. Limited monitoring scope and insufficient data fusion dimensions: Existing technologies mostly focus on single links such as docks and ships, lacking comprehensive coverage of all links in the entire process; no special monitoring system has been built for peak traffic characteristics, multi-source data has not achieved proactive perception and intelligent linkage, data processing lacks a dynamic credibility grading mechanism, and the evaluation results cannot adapt to the dynamic monitoring needs of peak periods.

[0004] 2. Lack of peak-period-specificity in weakness identification and resilience assessment: The lack of a full-chain linkage mechanism of "indicator anomaly - link weakness - causal root cause - resilience evolution" makes it difficult to accurately pinpoint the deep-seated causal root causes of weak resilience during peak periods; weaknesses can only be identified after the fact, lacking the ability to provide early warnings during peak periods, and it is impossible to combine multi-dimensional benefits to clarify the priority of improving weaknesses during peak periods.

[0005] 3. The assessment model is static and lacks core components: The existing assessment model has fixed parameters and does not consider the dynamic coupling effect between various links and the needs of peak periods, resulting in poor adaptability; it also does not build a peak period facility capacity assessment system for all links in the process, making it impossible to accurately measure the actual carrying capacity of core facilities during peak periods; the resilience assessment has a single dimension and does not cover the resilience evolution process during peak periods, making it difficult to quantify the system's comprehensive ability to cope with complex uncertainties.

[0006] 4. Incomplete scenario adaptation and lack of decision-making closed loop: No special evaluation system was designed for peak transportation periods, which cannot meet the differentiated needs of peak operation management; evaluation and decision-making are disconnected, no automated iterative closed loop is formed, optimization solutions are poorly implemented, and manual intervention is required for adjustment, making it difficult to achieve continuous autonomous improvement of the transportation system during peak periods.

[0007] Therefore, there is an urgent need for an integrated approach that combines intelligent perception throughout the entire process, multi-level causal root cause localization, quantification of facility capacity during peak periods across all transportation links, dynamic resilience evolution assessment, and autonomous iterative closed-loop optimization across the entire chain. This approach can overcome the shortcomings of existing technologies and meet the core needs of cross-strait passenger and vehicle ferry operations during peak periods. Summary of the Invention

[0008] To address the problems existing in the prior art, this invention provides a method and system for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and vehicle ferry transportation. The system focuses on constructing a comprehensive assessment framework for the entire process during peak transportation periods. This enables real-time and accurate perception of the operational status of the passenger and vehicle ferry transportation system, in-depth causal identification and early warning of weak links during peak periods, dynamic quantification of the full-process facility capabilities, dynamic evolution assessment of resilience levels across all scenarios, and autonomous generation, automatic execution, and continuous iteration of optimization schemes. It establishes a complete automated closed-loop system of "perception-identification-early warning-assessment-decision-execution-multi-dimensional feedback-autonomous optimization," providing comprehensive and high-precision scientific support for efficient operation of the transportation system, emergency dispatch during peak periods, risk prevention and control, green and low-carbon development, and resilience enhancement.

[0009] To achieve the above objectives, the present invention provides the following solution: A method for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off transportation, the method comprising: Collect multi-source data from the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to construct a five-dimensional dynamic relational database; Based on a five-dimensional dynamic relational database and a facility capacity coupled optimization evaluation model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, the peak-period facility capacity evaluation results are obtained. Based on the full-process peak-period facility capacity assessment results and the five-dimensional dynamic correlation database, the results of shortcoming location and root cause tracing are obtained through a hybrid root cause model. Based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to obtain the resilience assessment results. We construct a closed-loop process for autonomous iteration across the entire chain, driven by intelligent agents. Based on resilience assessment results, we enable the autonomous generation, execution, feedback, and iteration of optimization solutions during peak periods.

[0010] Preferred methods for intelligent preprocessing of collected multi-source data include: Edge computing technology is used to preprocess the collected multi-source data in real time, filter key data fields, remove invalid data and abnormal fluctuation values, and obtain filtered data. A data quality assessment model is constructed based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". The data quality assessment model is used to clean the screened data to obtain cleaned data. The cleaned data is repaired by using the methods of "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction" to obtain complete data; A four-dimensional dynamic weight model is constructed based on "data source credibility, data timeliness, scenario importance, and causal relationship strength" to obtain the weight values ​​of each supplementary data.

[0011] Preferred methods for obtaining peak-period facility capacity assessment results across all stages, based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, include: Port back-end collection and distribution channel capacity: ; in, : Peak-period capacity of port rear access and distribution channels; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor; : Mixed vehicle model correction factor; Average headway of vehicles; Average parking distance between vehicles; Channel scenario adjustment factor; Peak traffic flow correction factor; Route resource capabilities: ; in, Peak season route resource capacity; Number of shipping routes opened at the port; Maximum number of vehicles / passengers that a single ship can carry; Shortest departure interval during peak hours; : Coupling optimization function; Carbon footprint throughout its entire life cycle; Port resource capacity: Average daily departure capacity per berth: ; Total daily transport capacity of a single-sided wharf: ; in, Average loading and unloading time during peak hours at a single-sided wharf; Average berthing and departure time at peak times for a single-sided pier; Average interval between adjacent vessels during peak periods; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles a single vessel can carry; : Adjustment factor for berth scenarios; : Dock-ship dynamic coupling coefficient; Peak-hour terminal operation efficiency index; : Terminal-operation coupling optimization function; Shipping capacity: Average daily departure capacity of vessels: ; Total daily vehicle transport capacity of a single vessel: ; in, : No. Total transit time for a single vessel on this route; : No. Number of vessels operating on each route during peak periods; Number of routes; Ship scenario adjustment factor; : Ship-dock dynamic coupling coefficient; Peak-period ship operation efficiency index; Ship-operation coupling optimization function; Supporting facilities - service capacity of the rear gate: ; in, , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These represent the average time for trucks, passenger vehicles, and hazardous materials transport vehicles to clear customs during peak hours. : Gate scene adjustment factor; Energy consumption per unit of transport at the gate; Energy consumption weighting factor; Peak-hour gate operation efficiency index.

[0012] Preferably, the hybrid root cause model includes: an indicator layer, a link layer, a causal root cause layer, and an influence evolution layer; The indicator layer is used to identify abnormal indicators by comparing real-time monitoring data during peak periods with preset thresholds. The aforementioned process layer is used to trace the corresponding process weaknesses based on the correlation analysis of abnormal indicators. The causal root cause layer is used to identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithm and eliminate spurious correlations; then, the propagation path of the root cause is analyzed through FTA and the contribution of the root cause is quantified by combining Bayesian network. The impact evolution layer is used to combine the characteristics of peak transportation scenarios to predict the degree of impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios.

[0013] Preferably, based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, the method for obtaining resilience assessment results using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" includes: Dynamic toughness assessment: Peak resilience margin: ; Time-based dynamic adaptation coefficient: ; in, The overall carrying capacity of the system during peak hours; Actual transportation demand during peak hours; , These represent the overall system capacity and actual demand for each hour during peak periods; Disturbance resilience assessment: Disturbance immunity coefficient: ; Short-board disturbance transmission coefficient: ; in, , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. Capacity loss in bottleneck links during peak periods; Total system capacity loss during peak periods; Recovery assessment: Recovery efficiency coefficient: ; Contribution to emergency dispatch and recovery: ; in, The time it takes for the system to recover to normal capacity after a peak-period disturbance; Natural recovery time during peak periods without emergency dispatch; Resilience reserve capacity assessment: Capacity / Road Reserve Coefficient: ; Emergency resource response efficiency: ; in, Maximum carrying capacity of core facilities; The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; Synergistic resilience assessment: Cooperative response efficiency coefficient: ; Collaborative resilience enhancement coefficient: ; Intelligent early warning resilience assessment: ; Resilience evolution trend assessment: Resilience evolution rate: ; Toughness stability coefficient: ; in, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period; Peak-period comprehensive resilience index: ; in, , , , , , , : Assign optimal weights to the seven resilience dimensions respectively. , , , , , , : These are the standardized values ​​of each resilience index.

[0014] The present invention also provides an intelligent monitoring and dynamic resilience assessment system for the entire process of cross-strait passenger and roll-on / roll-off transportation. The system is used to implement the aforementioned method and includes: a data acquisition and processing module, a capacity assessment module, a bottleneck location and root cause tracing module, a resilience assessment module, and a closed-loop optimization module. The data acquisition and processing module is used to collect multi-source data of the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to build a five-dimensional dynamic relational database. The capacity assessment module is used to obtain the peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm. The shortcoming location and root cause tracing module is used to obtain the shortcoming location and root cause tracing results based on the full-process peak period facility capacity assessment results and the five-dimensional dynamic correlation database, through a hybrid root cause model. The resilience assessment module is used to obtain resilience assessment results based on the results of shortcoming location and root cause tracing and the seven-dimensional dynamic resilience assessment model, using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment". The closed-loop optimization module is used to construct a fully autonomous iterative closed-loop process driven by intelligent agents. Based on the resilience assessment results, it enables the autonomous generation, execution, feedback, and iteration of optimization schemes during peak periods.

[0015] Preferably, the intelligent preprocessing process for the collected multi-source data includes: Edge computing technology is used to preprocess the collected multi-source data in real time, filter key data fields, remove invalid data and abnormal fluctuation values, and obtain filtered data. A data quality assessment model is constructed based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". The data quality assessment model is used to clean the screened data to obtain cleaned data. The cleaned data is repaired by using the methods of "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction" to obtain complete data; A four-dimensional dynamic weight model is constructed based on "data source credibility, data timeliness, scenario importance, and causal relationship strength" to obtain the weight values ​​of each supplementary data.

[0016] Preferred methods for obtaining peak-period facility capacity assessment results across all stages, based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, include: Port back-end collection and distribution channel capacity: ; in, : Peak-period capacity of port rear access and distribution channels; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor; : Mixed vehicle model correction factor; Average headway of vehicles; Average parking distance between vehicles; Channel scenario adjustment factor; Peak traffic flow correction factor; Route resource capabilities: ; in, Peak season route resource capacity; Number of shipping routes opened at the port; Maximum number of vehicles / passengers that a single ship can carry; Shortest departure interval during peak hours; : Coupling optimization function; Carbon footprint throughout its entire life cycle; Port resource capacity: Average daily departure capacity per berth: ; Total daily transport capacity of a single-sided wharf: ; in, Average loading and unloading time during peak hours at a single-sided wharf; Average berthing and departure time at peak times for a single-sided pier; Average interval between adjacent vessels during peak periods; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles a single vessel can carry; : Adjustment factor for berth scenarios; : Dock-ship dynamic coupling coefficient; Peak-hour terminal operation efficiency index; : Terminal-operation coupling optimization function; Shipping capacity: Average daily departure capacity of vessels: ; Total daily vehicle transport capacity of a single vessel: ; in, : No. Total transit time for a single vessel on this route; : No. Number of vessels operating on each route during peak periods; Number of routes; Ship scenario adjustment factor; : Ship-dock dynamic coupling coefficient; Peak-period ship operation efficiency index; Ship-operation coupling optimization function; Supporting facilities - service capacity of the rear gate: ; in, , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These represent the average time for trucks, passenger vehicles, and hazardous materials transport vehicles to clear customs during peak hours. : Gate scene adjustment factor; Energy consumption per unit of transport at the gate; Energy consumption weighting factor; Peak-hour gate operation efficiency index.

[0017] Preferably, the hybrid root cause model includes: an indicator layer, a link layer, a causal root cause layer, and an influence evolution layer; The indicator layer is used to identify abnormal indicators by comparing real-time monitoring data during peak periods with preset thresholds. The aforementioned process layer is used to trace the corresponding process weaknesses based on the correlation analysis of abnormal indicators. The causal root cause layer is used to identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithm and eliminate spurious correlations; then, the propagation path of the root cause is analyzed through FTA and the contribution of the root cause is quantified by combining Bayesian network. The impact evolution layer is used to combine the characteristics of peak transportation scenarios to predict the degree of impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios.

[0018] Preferably, based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, the process of obtaining resilience assessment results using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" includes: Dynamic toughness assessment: Peak resilience margin: ; Time-based dynamic adaptation coefficient: ; in, The overall carrying capacity of the system during peak hours; Actual transportation demand during peak hours; , These represent the overall system capacity and actual demand for each hour during peak periods; Disturbance resilience assessment: Disturbance immunity coefficient: ; Short-board disturbance transmission coefficient: ; in, , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. Capacity loss in bottleneck links during peak periods; Total system capacity loss during peak periods; Recovery assessment: Recovery efficiency coefficient: ; Contribution to emergency dispatch and recovery: ; in, The time it takes for the system to recover to normal capacity after a peak-period disturbance; Natural recovery time during peak periods without emergency dispatch; Resilience reserve capacity assessment: Capacity / Road Reserve Coefficient: ; Emergency resource response efficiency: ; in, Maximum carrying capacity of core facilities; The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; Synergistic resilience assessment: Cooperative response efficiency coefficient: ; Collaborative resilience enhancement coefficient: ; Intelligent early warning resilience assessment: ; Resilience evolution trend assessment: Resilience evolution rate: ; Toughness stability coefficient: ; in, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period; Peak-period comprehensive resilience index: ; in, , , , , , , : Assign optimal weights to the seven resilience dimensions respectively. , , , , , , : These are the standardized values ​​of each resilience index.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention discloses an intelligent monitoring and dynamic resilience assessment method for the entire process of cross-strait passenger and vehicle roll-on / roll-off (Ro-Ro) transportation. By constructing a five-dimensional data intelligent linkage and fusion monitoring system, a full-process peak-season facility capacity coupling optimization assessment model, a multi-level dynamic assessment system for shortcomings, a seven-dimensional dynamic resilience improvement assessment system, and an intelligent agent-driven full-link autonomous iterative closed-loop process, it achieves real-time and accurate monitoring of the entire cross-strait passenger and vehicle Ro-Ro transportation process, deep-seated causal root cause localization of peak-season shortcomings, dynamic quantification of facility capacity, comprehensive assessment of resilience levels, and autonomous generation and iteration of optimization schemes. Compared with existing technologies, it has the following significant advantages: 1. Comprehensive data fusion dimensions significantly improve peak-period monitoring accuracy: An innovative five-dimensional dynamic correlation database of "ship behavior - operational status - natural environment - policy regulation - user behavior" is constructed. A three-level intelligent data processing mechanism and a four-dimensional dynamic weight allocation mechanism are designed for peak-period data characteristics to achieve deep linkage, intelligent cleaning and weighted fusion of multi-source data. The data completion accuracy rate reaches over 98%, and the data processing efficiency is improved by 60%. This solves the problems of limited monitoring scope, insufficient data fusion dimensions and inability to adapt to the dynamic needs of peak periods of existing technologies, and provides high-precision and highly dynamic data support for assessment.

[0020] 2. Accurate facility capacity assessment and significantly enhanced quantitative adaptability during peak periods: A peak-period facility capacity coupling optimization assessment model was constructed for the nine core links of cross-strait transportation. Dynamic coupling coefficients and peak-period-specific scenario adjustment factors were introduced, and algorithms such as NSGA-Ⅲ and DQN were used to optimize model parameters. The capacity quantification error of each link is ≤5%. The model accurately calculates the actual carrying capacity and capacity gap of core facilities during peak periods, solving the problems of static and fixed traditional models, failure to consider link coupling effects, and lack of peak-period specificity. This provides a scientific quantitative basis for the optimal allocation of resources.

[0021] 3. Deepening the identification and root cause analysis of shortcomings, and improving the efficiency of early warning and intervention during peak periods: By integrating causal inference, Bayesian networks and fault tree analysis, a multi-level shortcoming management system is constructed, consisting of "indicator anomalies - link shortcomings - causal root causes - impact evolution". The root cause identification accuracy is 95%. A dual-dimensional dynamic early warning module of "red-yellow-blue + trend" is established, with an early warning lead time of ≥4 hours. Combined with a four-dimensional cost-effectiveness model, improvement priorities are clarified, realizing early warning of shortcomings, hierarchical automated intervention and precise improvement. This solves the problems of ambiguous causal identification, lack of early warning and unclear improvement priorities of existing technologies during peak periods, and effectively avoids the spread of risks.

[0022] 4. Improved resilience assessment system, achieving a comprehensive breakthrough in quantifying system resilience during peak periods: A seven-dimensional dynamic resilience assessment model has been developed, oriented towards peak periods, encompassing "dynamic resilience, disturbance resistance resilience, resilience reserve, collaborative resilience, intelligent early warning resilience, and resilience evolution trend." This model includes core components such as port-backed collection and distribution channels within the assessment scope. Employing a scenario-based dynamic weight allocation mechanism, it comprehensively quantifies the system's resilience level and the resilience status of each dimension during peak periods. The overall resilience index matches the actual operational status with a 90% degree of accuracy. This overcomes the limitations of traditional resilience assessments, such as a single dimension, lack of peak-period focus, and incomplete coverage of the entire process, providing a clear direction for resilience improvement.

[0023] 5. Autonomous Iteration in the Decision-Making Closed Loop Significantly Improves the Intelligence of Peak-Hour Operation Management: A closed-loop autonomous iteration process driven by intelligent agents is constructed, encompassing the entire chain of "perception-evaluation-decision-execution-multi-dimensional feedback-autonomous optimization." This enables the autonomous generation, automatic execution, real-time feedback, and continuous iteration of optimization solutions during peak hours without human intervention. The response time for optimization solutions is ≤2 hours. Simultaneously, an intelligent interactive and visual decision support platform is developed to achieve multi-dimensional visualization and intelligent interactive simulation of evaluation results. This solves the problems of existing technologies being disconnected from evaluation and decision-making, reliant on manual intervention, and having poor implementation, significantly improving the efficiency and accuracy of peak-hour operation management.

[0024] 6. Multi-objective optimization and synergistic achievement, significantly improving the overall benefits of the transportation system: While improving transportation efficiency, risk control, and resilience during peak periods, the method of this invention incorporates a green and low-carbon assessment dimension, including indicators such as unit transportation energy consumption and carbon emissions into the facility capacity assessment and resilience assessment system. After the implementation of the optimized scheme, both transportation efficiency and green and low-carbon development can be achieved, while also improving user satisfaction and social benefits. This provides comprehensive scientific support for the efficient, safe, green, and resilient development of cross-strait passenger and vehicle roll-on / roll-off transportation, demonstrating outstanding novelty, creativity, and practicality, with broad application prospects. Attached Figure Description

[0025] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of a method for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off transportation, according to an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0029] Example 1 like Figure 1 As shown, this invention provides a method for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off transportation, including: Collect multi-source data from the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to construct a five-dimensional dynamic relational database; Based on a five-dimensional dynamic relational database and a facility capacity coupled optimization evaluation model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, the peak-period facility capacity evaluation results are obtained. Based on the full-process peak-period facility capacity assessment results and the five-dimensional dynamic correlation database, the results of shortcoming location and root cause tracing are obtained through a hybrid root cause model. Based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to obtain the resilience assessment results. We construct a closed-loop process for autonomous iteration across the entire chain, driven by intelligent agents. Based on resilience assessment results, we enable the autonomous generation, execution, feedback, and iteration of optimization solutions during peak periods.

[0030] The specific implementation process of this invention is as follows: 1. Construction of a Five-Dimensional Data-Driven Intelligent Linkage and Integration Monitoring System for the Entire Cross-Strait Passenger and Roll-on / Roll-off Transportation Process This data integrates AIS vessel big data (AIS data, navigation status data, etc., belonging to the "vessel behavior" dimension), port and shipping operation big data (terminal, vessel, and gate operation data, etc., belonging to the "operation status" dimension), meteorological and hydrological data (meteorological, sea state, and satellite remote sensing data, etc., belonging to the "natural environment" dimension), traffic flow monitoring data (covering traffic flow data across all stages of port rear access channels, hinterland access channels, and waiting areas, belonging to the "operation status" dimension), policy regulation data (route approval policies, holiday transportation control rules, and emergency response level standards, etc., belonging to the "policy regulation" dimension), and user behavior data (passengers). (Data such as cargo owner reservations, travel preferences, complaint feedback, and emergency response data are categorized under the "user behavior" dimension.) A comprehensive monitoring system is constructed covering the port's rear collection and distribution channels, hinterland collection and distribution channels, vehicle waiting areas, terminal scheduling, passenger and roll-on / roll-off ship transportation, onshore terminals and transportation, and supporting facilities. Specific monitoring indicators are set for different scenarios such as peak transportation periods, general transportation, dangerous goods transportation, emergency transportation, and extreme disturbance emergencies. The core monitoring indicators, definitions, and data sources for each link are clearly defined to achieve comprehensive perception of the entire process's operational status. This system is particularly well-suited for the accurate collection of data on peak traffic flow, facility operation, and emergency dispatch, as shown in Table 1.

[0031] Table 1 1.1 Construction of a Five-Dimensional Dynamic Relational Database Based on multi-source data types, a five-dimensional dynamic correlation database of "ship behavior, operational status, natural environment, policy regulation, and user behavior" is constructed to achieve intelligent linkage and dynamic updates of data in each dimension: when data in one dimension changes (such as a typhoon warning in the natural environment dimension or the introduction of peak-period transportation control rules in the policy regulation dimension), it automatically correlates data in other dimensions (such as navigation routes and speed thresholds in the ship behavior dimension, peak-period departure intervals and berth allocation rules in the operational status dimension, and peak-period travel demand adjustments in the user behavior dimension), and synchronously updates the calculation benchmarks of monitoring indicators and the parameters of the evaluation model to achieve dynamic adaptation of data to peak-period transportation scenarios.

[0032] 1.2 Data Intelligent Processing Mechanism In response to the characteristics of cross-strait transportation data, especially the large volume and high dynamism of data during peak periods, a three-level intelligent data processing mechanism is established to ensure data quality and the scientific nature of data fusion: Real-time stream preprocessing: Using edge computing technology, the collected data is preprocessed in real time at the data collection source, such as traffic flow monitoring equipment, ship AIS terminals, dock scheduling system, and IoT sensors. Key data fields are filtered, invalid data (such as AIS data during ship trial and sensor failure data) and abnormal fluctuation values ​​are removed, reducing the processing pressure on the cloud and improving data processing efficiency during peak periods.

[0033] Intelligent cleaning and completion: A data quality assessment model is built based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". It automatically identifies problems such as missing data, outliers, and spatiotemporal misalignment, accurately locates the causes of data anomalies, and achieves data repair through "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction". The completion accuracy rate is improved to over 98%, ensuring the integrity of monitoring data during peak periods.

[0034] Dynamic credibility weight allocation: Abandoning the fixed entropy weighting method, a four-dimensional dynamic weight model is constructed based on "data source credibility - data timeliness - scenario importance - causal correlation strength." This model calculates the weight of each data source in real time (e.g., in a typhoon warning and peak period scenario, the weight of meteorological satellite data automatically increases to 0.95, while the weight of manually recorded data decreases to 0.05; in a peak period hazardous materials transportation scenario, the weight of IoT monitoring data is 0.9, and the weight of dispatch system data is 0.85). Weighted fusion is performed during indicator calculation to improve the scientific nature and dynamic adaptability of data fusion. Specifically: During the facility capacity assessment phase: The core of the comprehensive peak-hour facility capacity coupling assessment is to quantify the capacity and load level of each link (collection and distribution channels, waiting areas, etc.), and the calculation of assessment indicators requires five-dimensional fused data as the core input. After the four-dimensional dynamic weight model calculates the weight values ​​of each data source, it will perform weighted fusion in the indicator calculation stage of the capacity assessment. That is, for the same assessment indicator, the data from different sources (such as the load indicator of collection and distribution channels, corresponding to traffic flow monitoring data and dispatch system data) are weighted and summed according to the weight values ​​to obtain accurate indicator quantification results, providing scientific data support for subsequent capacity bottleneck identification (such as in the peak-hour scenario in the original text, the weight of the operational status dimension data is increased, and its weighted fusion result directly affects the accuracy of core assessment indicators such as facility load and traffic efficiency).

[0035] In the auxiliary support stage: The identification of shortcomings and root cause analysis, as well as resilience assessment, all need to be based on the "weighted and integrated precise indicator data" output from the facility capacity assessment stage, without the need for separate weighted integration.

[0036] In the process of identifying shortcomings, the core of identifying "indicator anomalies" is to compare the actual value of the weighted and fused indicator with the warning threshold, thereby locating the shortcomings in the process and tracing the root cause (such as judging whether the traffic flow index exceeds the carrying capacity of the channel based on the weighted and fused traffic flow index). In resilience assessment, the calculation of the seven-dimensional resilience indicators requires precise indicator data output from the capability assessment as input. The weighted fusion has been completed in the capability assessment stage to ensure the scientific and accurate nature of the resilience assessment.

[0037] 2. Evaluation Model for Peak-Period Facility Capacity Coupling in Core Links of Cross-Strait Passenger and Roll-on / Roll-off Transportation Based on the intelligently preprocessed full-process monitoring data and combined with the operational patterns of each stage of cross-strait transportation, a multi-objective capacity coupling optimization evaluation model (i.e., facility capacity coupling optimization evaluation model) covering nine core stages is constructed. Focusing on peak-season transportation scenarios, scenario adjustment factors and dynamic coupling coefficients are set, and a non-dominated sorting genetic algorithm (NSGA-Ⅲ) is introduced to solve for the Pareto optimal solution, achieving accurate quantification and dynamic correction of facility capacity during peak seasons in each stage. This step provides a quantitative basis for identifying bottlenecks and allocating resources during peak seasons.

[0038] 2.1 Capacity Model of Port Back-End Collection and Distribution Channel Facilities ; In the formula: : Peak-hour capacity of port access transportation facilities (standard vehicles / hour), with standard vehicles converted to passenger cars as the standard unit; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor, including the effects of horizontal and vertical alignment, intersections, etc., with a value of 0.7~0.95; : Model mix correction coefficient, taken as 0.6~0.85 when the proportion of futures vehicles increases during peak periods; Average headway between vehicles (seconds / vehicle), taken as 1.8~2.5 seconds / vehicle during peak hours; Average parking distance per vehicle (meters / vehicle); The channel scenario adjustment factor is 0.08 for pure peak hours, 0.12 for peak hours combined with emergency transportation, and -0.15 for equipment maintenance / construction periods. Peak traffic flow correction factor, determined by combining the ratio of historical peak traffic flow to daily traffic flow, with a value ranging from 1.2 to 2.0.

[0039] 2.2 Route Resource Capacity Model (Coupled Optimization Model) Based on actual calculation methods for port and shipping operations, this paper integrates multi-objective optimization of resilience, efficiency, and green and low-carbon development, and uses the NSGA-Ⅲ algorithm to solve the Pareto optimal solution to meet the operational needs of routes during peak periods.

[0040] ; In the formula: Peak season route resource capacity (standard vehicles / day or people / day); Number of shipping routes opened at the port; Maximum number of vehicles / passengers that a single ship can carry; The shortest departure interval during peak hours (in minutes) is determined by combining the terminal's loading and unloading efficiency and ship navigation patterns. : Coupled optimization function, solved by NSGA-Ⅲ algorithm for "comprehensive resilience index ( - Transportation efficiency ( - Green and low-carbon level ( The Pareto optimal solution for ") takes values ​​from 0.8 to 1.05; Carbon footprint throughout the entire life cycle (including carbon emissions from the entire process of ship construction, operation, and scrapping); 2.3 Port Resource Capacity Model Formula basis: The single berth departure capacity is calculated based on the actual operation cycle of the terminal. The total capacity integrates the dynamic coupling effect of the terminal and the vessel and the adjustment of peak period scenarios. The coupling coefficient is optimized through reinforcement learning algorithm (DQN) to adapt to the terminal operation demand during peak periods.

[0041] Average daily departure capacity per berth: ; Total daily transport capacity of a single-sided wharf: ; In the formula: Average loading and unloading time (in hours) during peak hours at a single-sided wharf; The average berthing and departure time (in hours) during peak hours for a single-sided wharf is determined by combining actual port and shipping operation data. Average interval (in hours) between adjacent vessels during peak hours, taking into account safe operating distances and scheduling efficiency during peak hours; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles per vessel (standard vehicles / shift), refer to vessel operation approval standards; : The berth scenario adjustment factor is 0.1 for pure peak hours, 0.12 for peak hours combined with emergency transportation, and -0.1 for equipment maintenance periods; The dock-ship dynamic coupling coefficient is automatically learned through the DQN algorithm based on full-process monitoring data. It reflects the linkage between dock loading and unloading and ship departure, and its value ranges from -0.2 to 0.15. Peak-period terminal operation efficiency index, calculated by combining berth utilization rate and loading and unloading efficiency, with a value of 0.7~0.95; The terminal-operation coupling optimization function integrates the coupling coefficient and operational efficiency, and is solved using the NSGA-Ⅲ algorithm, with values ​​ranging from 0.85 to 0.98. 2.4 Ship carrying capacity model (DQN algorithm) The vessel departure capacity is calculated based on the route sailing cycle and the number of operating vessels. The total capacity integrates the dynamic coupling effect of vessels and terminals, and the scenario adjustment factor is adapted to different scenarios such as peak periods and weather disturbances. The sailing time is dynamically corrected in combination with sea conditions.

[0042] Average daily departure capacity of vessels: ; Total daily vehicle transport capacity of a single vessel: ; In the formula: : No. The total transit time (in hours) for a single vessel on each route is dynamically adjusted based on peak sea conditions, taking into account satellite remote sensing sea condition data. : No. Number of vessels operating on the route during peak periods (number of vessels); Number of routes; Adjustment factor for ship scenarios: -0.15 during periods of weather disturbance, 0.1 during periods of pure peak / emergency transport, and 0.06 during periods of green and low-carbon transport; : Ship-dock dynamic coupling coefficient, and Dynamic association, optimized in real time through the DQN algorithm, with a value range of [-0.15, 0.1]; Peak-period ship operation efficiency index, calculated by combining capacity utilization and punctuality rate, with a value of 0.75~0.95; The ship-operation coupling optimization function integrates the coupling coefficient and ship operating efficiency, and is solved using the NSGA-Ⅲ algorithm, with values ​​ranging from 0.8 to 0.95. The definitions of the remaining parameters are the same as in 2.3.

[0043] 2.5 Supporting Facilities - Service Capacity Model of Rear Gate The gate service capacity is calculated based on the number of gates for different vehicle types and the passage time, and integrates peak-hour scenario adjustments, energy consumption weights and operational efficiency to adapt to the operation characteristics of multiple vehicle types mixed at the gates during peak hours.

[0044] ; In the formula: , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These are the average times (in minutes) for trucks, passenger cars, and dangerous goods transport vehicles to clear customs during peak hours. Adjustment factor for gate scenarios: 0.05 for pure peak periods, 0.1 for peak periods combined with emergency transportation, and 0.04 for green and low-carbon periods; Energy consumption per unit of transportation at the gate (kWh / vehicle), calculated in real time based on IoT monitoring data; Energy consumption weighting factor, in green priority scenarios In a pure peak scenario, δ=0.1; Peak-hour gate operation efficiency index, calculated by combining gate throughput and equipment health status, with a value of 0.8~0.95.

[0045] 2.6 Other core capabilities Hinterland access and distribution facility capacity: The calculation formula is consistent with that of the port's rear access and distribution channels (2.1). The scenario adjustment factor and correction coefficient are adapted according to the characteristics of the hinterland roads, and the peak period is considered. .

[0046] Rear waiting and turnover capacity: based on a comprehensive calculation of the waiting area size, peak vehicle ratio, loading and unloading efficiency, emergency capacity margin, peak user reservation data, and collaborative scheduling effect.

[0047] On-shore terminal and transport capacity: Combining unloading efficiency and transport channel capacity, and referring to the calculation methods of terminal resource capacity (2.3) and collection and distribution channel capacity (2.1), the adjustment factor for peak period scenarios is 0.07.

[0048] 3. Dynamic monitoring and multi-level bottleneck assessment system for peak periods of cross-strait passenger and vehicle ferry transportation. Based on the peak-period facility capacity assessment results and real-time monitoring data throughout the entire process, a full-chain bottleneck management system is constructed, which includes "indicator anomaly identification - bottleneck location - multi-level causal root cause tracing - peak-period dynamic early warning - improvement priority ranking". The system focuses on the peak transportation period to achieve accurate location of weak links, in-depth causal root cause analysis, early warning and clear optimization priority, providing a basis for emergency dispatch and resource optimization during peak periods.

[0049] 3.1 Multi-level causal root cause localization model To address the bottlenecks in the transportation system during peak hours, a hybrid root cause model combining causal inference, multi-level Bayesian networks, and fault tree analysis (FTA) is constructed. This model delves into the root causes of bottlenecks during peak hours at four levels, eliminating spurious correlations and accurately pinpointing the core causes. The hybrid root cause model includes: indicator layer, link layer, causal root cause layer, and influence evolution layer; Indicator Layer (First Layer): By comparing real-time monitoring data during peak periods with preset thresholds, abnormal indicators (such as excessive congestion time in the port's rear access channels during peak periods, excessive waiting time for ferry crossings during peak periods, and excessively low utilization rate of terminal berths) are identified. Second-level process: Based on correlation analysis of abnormal indicators, trace the corresponding bottlenecks in the process (such as insufficient capacity of the port's rear collection and distribution channels, insufficient peak-hour expected ferry capacity, and low terminal loading and unloading efficiency). The specific process is as follows: 1. Core Model: A Bayesian Network (BN) model is adopted to construct an "abnormal indicator-linkage bottleneck" correlation network. The network nodes are defined as follows: the input nodes are "weighted and fused abnormal indicators" (such as traffic flow load indicators, terminal loading and unloading efficiency indicators, waiting area capacity utilization rate indicators, etc.), the output nodes are "each core operation link" (port back-end collection and distribution channels, peak waiting area, terminal loading and unloading, ship scheduling, etc.), the intermediate nodes are "correlation strength between indicators and links", and the edges between nodes represent causal relationships. The correlation strength is determined by the "causal relationship strength" output by the four-dimensional dynamic weight model mentioned above.

[0050] 2. Core formulas and functions: (1) Correlation strength calculation function: used to quantify the correlation between abnormal indicators and each link, and as the input of conditional probability between Bayesian network nodes. The formula is as follows: ; in: The correlation strength between the i-th abnormal indicator and the j-th link (value range 0~1); This is the weighted fusion value of the i-th abnormal indicator (derived from the weighted fusion result of the capability assessment stage); This is the quantized value of the operating status of the j-th stage; This represents the covariance between abnormal indicators and the operational status of the process. , These represent the variances of abnormal indicators and the operational status of each process, respectively. This is the weight value for "causal correlation strength" in the four-dimensional dynamic weight model, ensuring that the correlation strength calculation is adapted to the needs of the scenario.

[0051] (2) Formula for calculating the probability of a bottleneck in a process: Based on the posterior probability calculation of Bayesian networks, the probability of a bottleneck in each process is determined, and the formula is as follows: ; in: Let be the posterior probability that there is a weakness in the j-th link (the value ranges from 0 to 1). The conditional probability of all abnormal indicators occurring simultaneously when there is a weakness in the j-th link. The prior probability of a weakness in the j-th link (based on historical operational data statistics, with an initial value of 0.3~0.5 during peak periods). This represents the joint probability of all abnormal indicators occurring simultaneously.

[0052] 3. Specific execution steps: The first step is to input the "weighted fusion of abnormal indicator data" (such as excessive traffic flow load indicators, low terminal loading and unloading efficiency indicators, etc.) from the capacity assessment stage; the second step is to calculate the correlation strength using the correlation strength function. The process involves four steps: First, calculating the correlation strength between each abnormal indicator and each operational link, and selecting the indicator-link correspondence with a correlation strength ≥ 0.7 (i.e., strong correlation). Second, substituting the strong correlation and abnormal indicator data into the Bayesian network model, and calculating the posterior probability of a link having a shortcoming using the shortcoming probability calculation formula. Third, setting a probability threshold (e.g., 0.6), when the posterior probability of a link is ≥ 0.6, the link is determined to be a shortcoming link, and the specific shortcoming type is output (e.g., insufficient capacity of the port's rear collection and distribution channel, insufficient peak-hour expected ferry capacity, low terminal loading and unloading efficiency), thus completing the link-level shortcoming tracing.

[0053] Example: Traffic flow load index during peak hours ( If the weighted fusion value exceeds the warning threshold (abnormal), it is obtained through the correlation strength calculation function. (Strength of correlation with the port's rear collection and distribution channels) (Association strength with waiting areas), filter out strong associations ( -Port rear collection and distribution channel); After substituting into the Bayesian network calculation, the posterior probability of this link having a short link is 0.78≥0.6, so it is determined to be a short link, and the short link is traced back to "insufficient traffic capacity of the port rear collection and distribution channel".

[0054] Causal root cause layer (third layer): Identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithms (potential outcome model, structural causal model) and eliminate spurious correlations; then analyze the propagation path of the root cause through FTA and combine Bayesian network to quantify the contribution of the root cause (e.g., the direct causal root cause of congestion in the port's rear collection and distribution channel is insufficient number of lanes, and the indirect causal root cause is inefficient vehicle classification and scheduling during peak hours). Impact Evolution Layer (Fourth Layer): Combining the characteristics of peak-hour transportation scenarios, predict the impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios (peak duration, peak combined with extreme weather, peak combined with policy changes). (For example, congestion in the port's rear access channels during peak hours leads to vehicle backlog at the waiting area, which in turn causes a decline in terminal operation efficiency, ultimately resulting in a continuous decrease in system resilience.) The specific process is as follows: 1. Core Model: The system dynamics (SD) model is used as the core analysis tool, combined with the Markov chain prediction model, to construct a three-dimensional evolution model of "short-terminal bottleneck - scenario disturbance - system resilience". The core is to simulate the diffusion process and resilience evolution trend of the bottleneck under different scenarios by quantifying the transmission path and feedback mechanism of the bottleneck's impact. This makes up for the limitations of traditional predictions that "emphasize static and neglect dynamic". It is suitable for the evolution prediction needs of multiple scenarios during peak periods (peak duration, peak superimposed with extreme weather, peak superimposed with policy changes).

[0055] 2. Core formulas and functions: (1) Formula for calculating the diffusion intensity of the impact of the shortest link: The degree of impact of the shortest link in a quantitative link on other related links is used as the basic input for evolution prediction. The formula is as follows: ; in: The influence diffusion intensity of the j-th link's weakness on the k-th related link at time t (value range 0~1). Let be the initial influence intensity of the j-th bottleneck (calculated from the posterior probability of the bottleneck). (i.e., the posterior probability of the weakest link calculated by the Bayesian network mentioned above). The diffusion coefficient (values ​​range from 0.02 to 0.08, with 0.08 for peak conditions combined with extreme weather, and 0.03 for normal peak conditions). For time variables (unit: h, representing the duration of the shortest board); The correlation coefficients for links j and k are derived from historical data statistics, such as the correlation coefficient between the port's rear collection and distribution channel and the waiting area for ferry crossings being 0.85, and the correlation coefficient between the wharf and ship transportation being 0.92. This is a scene correction function, with different values ​​corresponding to different scenes: peak duration period Peak season coincides with extreme weather Peak coincides with policy changes .

[0056] (2) System resilience evolution trend function: Based on the Markov chain model, the dynamic evolution trend of system resilience under different scenarios is predicted, and the formula is as follows: ; in: The system resilience index at time t+1 (the value ranges from 0 to 1, and the higher the index, the stronger the resilience). The initial system resilience index at time t (derived from the seven-dimensional dynamic resilience assessment results mentioned above); The Markov chain state transition probability matrix (i represents the current resilience state, j represents the next resilience state, trained based on historical peak data, covering three states: "strong resilience - medium resilience - weak resilience"). The average influence diffusion intensity of each related link at time t (from The weighted average is used to obtain the result, where the weights are the correlation coefficients of the links. ); This is the toughness attenuation coefficient (values ​​range from 0.05 to 0.12, with 0.12 for extreme weather scenarios).

[0057] (3) Function for determining the diffusion path of the impact of the shortest board: used to determine the transmission order and path of the impact of the shortest board, the formula is as follows: ; in: The optimal diffusion path for the bottleneck in the j-th link (i.e., the transmission path of the related links with the most significant impact); To find the maximum value function, the most significant transmission path is selected by comparing the "correlation coefficient × influence diffusion intensity" of each related link.

[0058] 3. Specific execution steps: The first step is to input basic data: trace the "short-board link" and "initial influence intensity of the short-board" from the link-level short-board analysis. "Combining the correlation coefficients of each stage" Scene correction functions for different scenarios The first step is to use the shortest path as the model input parameter; the second step is to calculate the diffusion intensity: using the formula for calculating the diffusion intensity of the shortest path's influence, the influence intensity of the shortest path on each related link at different times (t=1,2,...,n) is calculated. The third step is diffusion path determination: using a path determination function, the core diffusion paths of the impact of the bottleneck are screened out, and the transmission sequence is clarified (e.g., congestion in the port's rear access channels → vehicle backlog at the waiting area → decreased terminal operation efficiency). The fourth step is resilience evolution prediction: using a system resilience evolution trend function, combined with a Markov chain state transition probability matrix, the system resilience index for different time periods in the future is predicted. The fifth step is to output the results: output the degree of impact, diffusion path, and resilience evolution curve of the bottleneck in different scenarios, and clarify the scope and duration of the bottleneck's continuous impact on system operation.

[0059] Example: In a peak-period extreme weather scenario with heavy fog, the port's rear access road (j=1) has insufficient capacity. The initial impact intensity... (Converted from the posterior probability of 0.78), diffusion coefficient Scene correction function The correlation coefficient with the waiting field (k=1) Correlation coefficient with dock operations (k=2) Substituting into the diffusion intensity formula, at t=6h, , ; derived from the path determination function (Collection and distribution channel → dock operation) is the core diffusion path; substituting into the resilience evolution function, if the initial resilience index transition probability Average influence intensity attenuation coefficient ,but The prediction indicates a continuous decline in system resilience, consistent with the evolutionary trend mentioned by the agent: "Congestion in the port's rear access channels during peak periods leads to a backlog of vehicles waiting at the ferry terminal, which in turn causes a decrease in terminal operation efficiency, ultimately resulting in a continuous decline in system resilience."

[0060] 3.2 Peak-period bottleneck dynamic early warning module In response to the characteristics of peak-season transportation—high volume, high timeliness requirements, and easy spread of risks—a dual-dimensional dynamic early warning module based on "red-yellow-blue + trend" was established. This module includes the port's downstream collection and distribution channels within the core early warning scope, enabling early warning and tiered intervention for bottlenecks. Warning threshold setting: Based on historical operation data of the entire process, actual monitoring data during peak periods, and operation targets of the transportation system, a three-color warning threshold of "red-yellow-blue" is set for the facility capacity indicators during peak periods of each link. At the same time, a "peak period trend warning threshold" is set in combination with the resilience evolution trend (e.g., when the capacity of the port's rear collection and distribution channel is close to 80% of the peak period demand and shows a downward trend, a blue warning is triggered; when it is close to 70%, a yellow warning is triggered; and when it is below 60%, a red warning is triggered). Real-time early warning push: By comparing peak period monitoring data with early warning thresholds in real time, and combining resilience evolution trend analysis, the corresponding level of early warning is automatically triggered. Early warning information is pushed to managers through visualization platforms, SMS, system pop-ups, etc., to clarify abnormal indicators, related links, potential impacts and causal roots, and ensure the timeliness of early warning information. Early warning and coordinated intervention: Based on the early warning level during peak periods, tiered automated intervention measures are set up to achieve rapid response to shortcomings. Blue alert: The intelligent agent will automatically initiate minor optimization measures (such as adjusting the number of gates open during peak hours, optimizing vehicle routes in the port's rear collection and distribution channels, and adjusting departure intervals during peak hours). Yellow alert: Activate the moderate intervention plan (such as calling up backup equipment, increasing staffing during peak periods, and opening the emergency lane of the port's rear collection and distribution channel). Red Alert: The emergency dispatch system is activated to implement emergency plans (such as calling up backup vessels, activating temporary waiting areas, opening emergency transport channels during peak periods, and adjusting traffic control rules for port rear collection and distribution channels).

[0061] 3.3 Focusing on the shortcomings in peak and complex scenarios The focus is on peak transportation periods, while also covering complex scenarios involving peak and extreme conditions, as well as hazardous materials transportation. Specialized efforts are being made to address specific weaknesses and accurately assess capacity gaps and resilience losses. Peak-hour bottlenecks: For peak transportation periods (traffic volumes 2 times or more of daily averages), calculate the capacity gaps and resilience degradation rates of facilities at each stage during peak hours. ; ; In the formula: Actual transportation demand during peak hours (adjusted based on user reservation data); Maximum carrying capacity during peak hours, including the capacity of the port's rear collection and distribution channels; System resilience index at the initial stage of peak; System resilience index after the peak lasts for t hours; Peak resilience evolution decay rate (% / h). Capacity gap during peak hours.

[0062] Focusing on the shortcomings of peak and extreme scenarios: Based on the Agent-Based Modeling (ABM) method, ships, passengers, terminals, collection and distribution channels, emergency resources, etc. are regarded as intelligent agents. A cross-strait exclusive composite scenario (typhoon + strong ocean current + peak period, heavy fog + dangerous goods transportation + peak period) is constructed to simulate the interaction behavior of each intelligent agent under the composite scenario and to measure the operational status, capacity gap, resilience loss and evolution trend of each link, including the collection and distribution channels behind the port. Focusing on Shortcomings in Peak and Special Scenarios: For situations where peak periods overlap with special scenarios such as centralized transportation of hazardous materials and green and low-carbon transportation, calculate the gaps in specific indicators and their causal roots (such as the gap in the safety guarantee capacity of hazardous materials transportation in port rear collection and distribution channels during peak periods, and the gap in energy consumption control targets during peak periods).

[0063] 3.4 Priority Assessment of Shortcomings During Peak Periods A four-dimensional cost-effectiveness assessment model is introduced, encompassing "lifecycle cost, environmental impact, social benefits, and operational benefits." This model, combined with the core needs of peak-season transportation (efficiency, safety, and emergency response), calculates the cost-effectiveness of addressing shortcomings and prioritizes improvements based on cost-effectiveness, ensuring the scientific validity and feasibility of peak-season improvement plans. ; In the formula: : The improvement value of peak-season resilience index after the improvement of the weak link; The social benefits of the improved plan (such as reducing the number of passengers stranded during peak periods, reducing the probability of traffic accidents during peak periods, and alleviating congestion in the port's rear access channels during peak periods). Improve the operational efficiency of the plan (such as increasing peak-hour transportation efficiency and reducing peak-hour operating costs); Incremental environmental impact (such as increased energy consumption and emissions from newly added channels / berths); The total lifecycle cost of the solution (construction + operation + maintenance + disposal).

[0064] At the same time, a knowledge base for improvement solutions during peak periods is built. Based on machine learning algorithms, it associates historical peak period shortcomings, causal root causes, and optimal solutions, automatically matches customized improvement paths, and verifies the implementation effect and resilience evolution improvement through full-process monitoring data.

[0065] 4. A seven-dimensional dynamic resilience enhancement assessment system oriented towards peak periods of cross-strait passenger and vehicle ferry transportation. By integrating peak-period bottleneck identification, causal root cause tracing, and full-process monitoring data, a seven-dimensional dynamic resilience assessment model is constructed, comprising "dynamic resilience, disturbance resistance resilience, resilience reserve, collaborative resilience, intelligent early warning resilience, and resilience evolution trend." All indicators are designed specifically for peak-period transportation scenarios, distinguishing between different transportation types and complex scenarios. A dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to achieve comprehensive quantification of the system's resilience level during peak periods and scientific evaluation of resilience improvement effects.

[0066] 4.1 Dynamic toughness assessment Quantifying the match between system facility capacity and transportation demand during peak periods reflects the basic resilience level of the system during peak periods: Peak resilience margin: ; Time-based dynamic adaptation coefficient: ; In the formula: During peak hours, the system's comprehensive carrying capacity integrates the core capabilities of port rear access channels, wharves, and ships. Actual transportation demand during peak hours (adjusted based on user reservation data); , These represent the overall system capacity and actual demand for each hour during peak periods.

[0067] 4.2 Disturbance Resistance Assessment Assess the system's resilience to interference during peak periods under the influence of factors such as weather disturbances, equipment failures, and traffic congestion, as well as the degree of disturbance transmission in weak links (especially the port's downstream transport channels): Disturbance immunity coefficient: ; Short-board disturbance transmission coefficient: ; In the formula: , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. : Capacity loss in bottleneck links (such as port rear collection and distribution channels) during peak periods; Total system capacity loss during peak periods.

[0068] 4.3 Resilience Assessment The efficiency of the system's recovery to normal capacity after being disturbed during peak periods is measured, as well as the contribution of emergency dispatch measures to the recovery process, reflecting the resilience of emergency response during peak periods. Recovery efficiency coefficient: ; Contribution to emergency dispatch and recovery: ; In the formula: : The time (in hours) for the system to recover to normal capacity after a peak period disturbance, including the recovery time of the port's rear collection and distribution channels; The natural recovery time during peak periods without emergency dispatch is predicted based on historical data and simulation models.

[0069] 4.4 Resilience Reserve Capacity Assessment Assessing the redundancy capacity of core facilities such as ships, docks, and port access routes during peak periods, as well as the response speed of emergency resources, is an important reserve indicator for enhancing resilience during peak periods. Capacity / Road Reserve Coefficient: ; Emergency resource response efficiency: ; In the formula, : Maximum carrying capacity of core facilities (ship / terminal capacity, port rear access channel capacity); The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; 4.5 Collaborative Resilience Assessment The overall coordination capability of "port rear collection and distribution channels - hinterland collection and distribution channels - waiting area - wharf - ships" during peak periods is quantified, and the effect of coordination measures on improving system resilience is demonstrated, reflecting the interconnected resilience of all links in the entire process: Cooperative response efficiency coefficient: ; Collaborative resilience enhancement coefficient: ; Where: Coordinated response completion time: the total time spent by all links in the entire process to respond to disturbances during peak periods (including the coordinated scheduling time of the port's rear collection and distribution channels); Average independent response time: the average time spent by each link to respond to disturbances individually during peak periods, the lower the value, the stronger the coordination.

[0070] 4.6 Intelligent Early Warning Resilience Assessment The effectiveness of the peak-period early warning mechanism and the improvement in system resilience achieved by optimization measures within the early warning time window are evaluated to reflect the "early warning-optimization" logic for enhancing peak-period resilience. ; Where: Warning duration: the time window (in hours) from the triggering of the warning to the occurrence of the disturbance during the peak period; Post-warning resilience index: the system resilience index after optimization measures are taken within the warning time window during the peak period (including the optimization effect of the port's rear collection and distribution channels); Pre-warning resilience index: the original system resilience index when the warning is not triggered during the peak period.

[0071] 4.7 Assessment of Resilience Evolution Trend Monitoring the real-time trends and stability of system resilience during peak periods provides a basis for the dynamic adjustment of resilience enhancement measures during peak periods. Resilience evolution rate: ; Toughness stability coefficient: ; In the formula, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period.

[0072] 4.8 Calculation of Comprehensive Resilience Index A Bayesian optimization algorithm combined with scenario risk clustering is used to achieve dynamic weight allocation, and the optimization is adapted to the characteristics of passenger and roll-on / roll-off transportation scenarios during peak periods.

[0073] The algorithm employs "Bayesian optimization + scenario risk clustering + causal weight adjustment" to automatically identify peak-period related scenario types (such as pure peak scenarios, peak + heavy fog scenarios, peak + hazardous materials transportation scenarios) and assign optimal weights to the seven-dimensional resilience dimension. , , , , , , ), and combine the standardized index values ​​to calculate the peak-period comprehensive resilience index: ; In the formula, , , , , , , : These are the standardized values ​​of each resilience index, processed using the min-max standardization method, with values ​​ranging from [0,1]. Example of dynamic adjustment for peak options: Peak hours: , , , , , , ; Peak hours combined with heavy fog: , , , , , , ; Peak hours + dangerous goods transportation period: , , , , , , .

[0074] 5. Intelligent agent-driven end-to-end autonomous iterative closed-loop evaluation and optimization decision-making process Based on five-dimensional data monitoring, full-process peak-season facility capacity assessment, multi-level bottleneck identification, and seven-dimensional resilience assessment results, a fully autonomous iterative closed-loop process driven by an intelligent agent is constructed. This intelligent agent (which is an intelligent decision-making and execution module adapted to peak seasons and complex scenarios of cross-strait passenger and vehicle ferry transportation, autonomously receiving multi-source data and assessment results, and completing the entire process of optimization scheme generation, execution, feedback, and iteration without human intervention; it is the core driving unit of the "perception-assessment-decision-execution-multi-dimensional feedback-autonomous optimization" closed-loop process) focuses on achieving the autonomous generation, execution, feedback, and iteration of peak-season optimization schemes, forming an automated closed loop of "perception-assessment-decision-execution-multi-dimensional feedback-autonomous optimization" without human intervention, improving the efficiency and accuracy of peak-season operation management. The core processes include: Multi-source data collection: Data on special monitoring indicators for the entire process and peak periods are collected through traffic flow monitoring equipment, AIS ship positioning system, terminal scheduling and management system, policy release platform, user behavior collection terminal, IoT sensor, satellite remote sensing equipment, etc., with a focus on collecting operational data of the port's rear collection and distribution channels; Intelligent data preprocessing: Real-time stream preprocessing based on edge computing, data cleaning, completion and spatiotemporal registration are completed through a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer", and a five-dimensional dynamic relational database is constructed by combining a four-dimensional dynamic weight model to allocate data credibility weights. Peak period facility capacity assessment: Substitute the pre-processed data into the core link peak period facility capacity coupling optimization assessment model, and combine the dynamic coupling coefficient and peak period scenario adjustment factor to calculate the peak period facility capacity value of each link, including the port rear collection and distribution channel. Peak-period multi-level bottleneck location and early warning: Based on the results of link capability assessment and real-time monitoring data of the whole process, the deep-root causes of bottlenecks during peak periods, including the port's rear collection and distribution channels, are located through the "causal inference-Bayesian network + FTA" model, triggering corresponding level early warnings and pushing automated intervention suggestions. Seven-dimensional dynamic resilience assessment: Combining the results of peak period weakness identification and causal root cause tracing, the results are substituted into the seven-dimensional dynamic resilience assessment model. The dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to calculate the peak period comprehensive resilience index, specific resilience index and resilience evolution trend. Intelligent Agent Autonomous Optimization Decision Generation and Verification: Based on the peak period bottleneck priority, resilience assessment results, causal root causes and improvement scheme knowledge base, the intelligent agent autonomously generates peak period exclusive optimization schemes, focusing on optimization measures for port back-end collection and distribution channels (such as adjusting channel traffic organization, opening emergency lanes, and optimizing vehicle scheduling). It also includes resource allocation adjustments, emergency dispatch strategies, and facility improvement suggestions for wharves, ships, waiting areas, and other links. The implementation effect, comprehensive benefits and resilience evolution improvement of each scheme are verified through full-process monitoring data. Automatic execution of optimization plans: Seamlessly integrates with port collection and distribution management systems, terminal scheduling systems, ship operation management systems, traffic control systems, etc., and automatically distributes verified peak-hour optimization plans to the corresponding systems for execution; Multi-dimensional real-time feedback: The data acquisition system monitors the implementation effect of the optimization plan in real time, and compares the changes in peak-period process capabilities, bottleneck status, causal root causes, resilience index, resilience evolution trend, user satisfaction, equipment health status, operational efficiency and green and low-carbon benefits before and after the implementation of the plan. Intelligent agent autonomous iterative optimization: Based on multi-dimensional feedback data, the intelligent agent automatically adjusts the evaluation model parameters, dynamic weights and optimization schemes to form a closed loop of autonomous iterative optimization across the entire chain during peak periods; Results Output and Visualization: Generate peak period special monitoring reports, shortcoming analysis reports, causal root cause analysis reports, resilience assessment reports, and optimization decision reports. The visualization platform intuitively displays the entire process operation status, the distribution of shortcomings during peak periods, including port rear collection and distribution channels, causal root cause maps, resilience index trends, resilience evolution curves, and the implementation effects of decision-making schemes.

[0075] 6. Construction of an intelligent interactive and visual decision support platform Develop an intelligent interactive visualization platform for resilience assessment of cross-strait passenger and roll-on / roll-off (Ro-Ro) transportation, adding a special monitoring and assessment module for port back-end collection and distribution channels. Focusing on peak-season functions, the platform integrates core features such as full-process monitoring, assessment result display, intelligent decision-making, and scenario simulation. It provides managers with visualized and intelligent peak-season operational decision support. Core functions include: Real-time monitoring of the entire process: Using a map as a medium, the system visualizes the congestion of the port's rear access channels during peak hours, the flow of the hinterland access channels, the distribution of vehicles in the waiting area during peak hours, the navigation position of ships, the occupancy of berths during peak hours, and the dispatch trajectory of emergency resources during peak hours, allowing for an intuitive understanding of the system's operating status. The assessment results are visualized in multiple dimensions: radar charts are used to display the peak-period facility capacity, the distribution of shortcomings, the peak-period resilience index trend curve, the causal root cause analysis map, and the seven-dimensional resilience radar chart of each link, including the port rear collection and distribution channel, making the assessment results easier to understand and apply. Intelligent Interaction and Peak Season Solution Simulation: Supports natural language interaction (voice / text input requirements), intelligent agent automatically generates peak season-specific assessment reports; supports users to manually adjust parameters (such as optimizing the number of lanes in the port's rear collection and distribution channels, modifying peak season departure intervals, and adjusting the number of gates open during peak seasons) for real-time simulation, allowing users to intuitively view the impact of the solution on system resilience, efficiency, and cost. Role-based dynamic personalization: Based on user behavior data and peak season scenario needs, the service content and display format are adjusted in real time for different roles such as port and shipping management departments, transportation companies, emergency command centers, passengers / cargo owners, etc., and exclusive evaluation reports and function interfaces are automatically generated. Peak period custom scenario simulation: Allows users to upload custom peak period composite scenario parameters (such as "extreme typhoon + sudden 3 times passenger flow + peak period" or "concentrated transportation of dangerous goods vehicles + heavy fog + peak period"). The system automatically generates simulation scenarios and completes the evaluation and solution output of the whole process, including the port rear collection and distribution channel. Peak period early warning information and execution status push: Real-time push of early warning information on bottlenecks in the port's rear collection and distribution channels during peak periods, execution progress of optimization plans, alarms on abnormal situations, etc., supporting remote intervention and decision adjustment by management personnel; Facility Capacity and Resilience Trend Tracking: Real-time tracking of the changing trends of facility capacity and resilience indices during peak periods at each stage, highlighting the optimization effects of port back-end collection and distribution channels, and providing data support for subsequent operation and management.

[0076] In summary, this invention provides a method for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off transportation, which has the following advantages compared with existing technologies: 1. Five-dimensional data intelligent linkage and integration for precise adaptation to peak period dynamic monitoring: Construct a five-dimensional dynamic correlation database of "ship behavior - operational status - natural environment - policy regulation - user behavior", and design a four-dimensional dynamic weight allocation mechanism for peak period data characteristics to achieve deep linkage and intelligent processing of multi-source data. This solves the problems of insufficient data fusion dimensions and poor real-time performance in existing technologies, and provides high-precision and high-dynamic data support for peak period assessment.

[0077] 2. Construct a peak-period facility capacity coupling optimization model to achieve accurate quantification: Construct a peak-period facility capacity coupling optimization evaluation model for each core link of cross-strait transportation (including port back-end collection and distribution channels), introduce dynamic coupling coefficients and peak-period-specific scenario adjustment factors to achieve accurate quantification of peak-period facility capacity, and solve the problems of static solidification and failure to adapt to peak periods in traditional models.

[0078] 3. Multi-level dynamic assessment of bottlenecks during peak periods to achieve accurate early warning and root cause localization across all stages: By integrating causal inference, Bayesian networks and fault tree analysis, a peak-period-specific bottleneck management system covering the port's rear collection and distribution channels is constructed. This enables in-depth causal root cause mining of bottlenecks and dynamic early warning based on "red-yellow-blue + trend" indicators. Combined with a four-dimensional cost-effectiveness model, improvement priorities are clearly defined, solving the problems of ambiguous causal identification and lack of early warning for bottlenecks during peak periods.

[0079] 4. Build a peak-season-oriented seven-dimensional resilience assessment system to quantify the resilience improvement effect of the whole process: Construct a seven-dimensional dynamic resilience assessment model, incorporate the resilience status of the port's back-end collection and distribution channels into the core assessment scope, design all indicators for peak seasons, and adopt a scenario-based dynamic weight allocation mechanism to achieve a comprehensive quantification of the system's resilience level during peak seasons, breaking through the limitations of traditional resilience assessments that lack peak-season-specificity and full-process coverage.

[0080] 5. Intelligent agent-driven end-to-end autonomous iteration closed loop improves peak-hour operational efficiency: Establish a closed-loop system of "intelligent agent autonomous decision-making + end-to-end autonomous iteration" to realize the autonomous generation, automatic execution and continuous iteration of peak-hour optimization solutions. It focuses on the emergency optimization of port rear collection and distribution channels. Combined with the peak-hour special functions of the visualization platform, it significantly improves the practical guidance and implementation of the technology and solves the problems of disconnect between assessment and decision-making and reliance on manual labor.

[0081] Through the aforementioned core innovations, this invention effectively compensates for the shortcomings of existing technologies, focuses on solving the core pain points of cross-strait passenger and vehicle ferry operations during peak periods, improves the whole-process evaluation system, and provides a brand-new integrated technical solution for efficient operation, peak-period emergency dispatch, risk prevention and control, green and low-carbon development, and resilience enhancement of the transportation system, with broad application prospects.

[0082] Example 2 The intelligent monitoring and dynamic resilience assessment method for the entire process of cross-strait passenger and vehicle ferry transportation of the present invention is applicable to various cross-strait passenger and vehicle ferry transportation scenarios. It conducts peak-season operation management and assessment for all aspects, including the port's rear collection and distribution channels. This embodiment combines the general engineering scenario of cross-strait passenger and vehicle ferry transportation during the Spring Festival travel rush to provide a detailed description of the method described in the aforementioned embodiment, and to verify the scientificity, effectiveness and engineering applicability of the method.

[0083] 1. Multi-source data acquisition and intelligent preprocessing Data was collected on cross-strait passenger and roll-on / roll-off transportation during the peak Spring Festival travel season (40 days in total), covering 9 core stages of the entire process. Data types included: Data on port rear access and distribution channels: 4 effective lanes, average daily traffic volume during peak hours is 38,000 vehicles, traffic speed is 25km / h, average daily congestion duration is 1.5 hours, average daily emergency transport channel has 30 vehicles, and IoT energy consumption monitoring data shows that the unit transport energy consumption is 0.5kWh / vehicle; AIS vessel data: 5955 representative voyages were screened to extract peak-period operational data such as vessel berthing time, sailing speed, punctuality rate, and capacity utilization rate; Port and shipping operation data: The port area has 10 passenger and vehicle roll-on / roll-off berths in operation, with a normal daily departure of 90 trips and a maximum daily departure of 102 trips. The peak daily transport demand is 23,400 vehicles, and an average of 120 dangerous goods transport vehicles pass through each day. The average loading and unloading time during peak periods is 1.2 hours, and the berthing and departure time is 0.5 hours. Meteorological, hydrological, and satellite remote sensing data: Recorded two fog disturbance events (each lasting 2-3 hours) and one typhoon warning event (lasting 4 hours), and obtained data such as wave height and ocean current speed to provide a basis for adjusting ship navigation time during peak periods; Traffic flow and IoT data: The average daily traffic volume on the inland expressway is 43,000 vehicles, the average daily turnover of the port waiting area is 18,000 vehicles, the good health rate of the terminal loading and unloading equipment is 98%, and the gate is divided into 4 trucks, 6 passenger cars, and 2 dangerous goods. Policy control data: During the Spring Festival travel rush, the emergency response level is Level II, and dangerous goods transport vehicles are prohibited from passing through from 22:00 to 6:00 the next day, while emergency transport vehicles are given priority during peak periods. User behavior data: Peak passenger bookings are concentrated between 9:00-11:00 and 14:00-16:00 daily. The preference rate for core routes is 65%, the waiting satisfaction rate is 82%, and the complaint hotspots are congestion in the port's rear access channels and excessive waiting time.

[0084] Edge computing technology was used for real-time data preprocessing. A hybrid model combining attention mechanism, Transformer, and spatiotemporal feature transfer was used to fill in 23 missing data points and correct 51 abnormal fluctuation values, achieving a data completion accuracy of 98.6%, which meets the data integrity requirements for peak-period monitoring. A four-dimensional dynamic weight model was used to determine the weight of each data source (in the typhoon warning + peak-period scenario, meteorological satellite data has a weight of 0.95, port and shipping operation system data has a weight of 0.9, and manually recorded data has a weight of 0.05; in the peak-period port rear collection and distribution channel scenario, IoT monitoring data has a weight of 0.9, and traffic dispatch system data has a weight of 0.85), completing the weighted fusion of multi-source data and constructing a five-dimensional dynamic relational database of "ship behavior - operational status - natural environment - policy regulation - user behavior", ensuring the quality and scientific rigor of peak-period data fusion.

[0085] 2. Peak-period facility capacity assessment The preprocessed data was substituted into the peak-period facility capacity coupling optimization and evaluation model of each core component. The NSGA-Ⅲ algorithm was used to solve for the Pareto optimal solution, and the dynamic coupling coefficient was optimized by combining the DQN algorithm to complete the accurate quantification of the peak-period facility capacity of each component. The calculation results of the core components are as follows: 2.1 Port Back-End Collection and Distribution Channel Facilities Capacity According to the formula Calculation, parameter values: , (Lane width 3.75m) (Including 2 intersections) (During peak season, futures vehicles accounted for 40%). Vehicles, Vehicles, (Pure peak hours) (The peak traffic volume during the Spring Festival travel rush is 1.8 times that of the daily average).

[0086] The calculated throughput is 1,865 standard vehicles per hour, with a daily average capacity of approximately 44,800 standard vehicles. Combined with the actual daily average traffic flow of 38,000 vehicles, the theoretical carrying capacity of the channel meets the demand, but congestion actually exists. This is determined to be an efficiency loss caused by insufficient optimization of the scheduling strategy.

[0087] 2.2 Port Resource Capacity The formula for the average daily departure capacity per berth Calculation, parameter values: , , (Interval between adjacent vessels during peak periods) (During peak hours), the calculated number of berths is 13.2 per day.

[0088] The total daily transport capacity of a single-sided wharf is based on the formula. Calculation, parameter values: , Standard vehicle / shift (Dock-ship dynamic coupling coefficient, optimization result of DQN algorithm). (Peak Terminal Operation Efficiency Index). (Results from NSGA-Ⅲ algorithm) The calculated value is 14,784 standard vehicles per day.

[0089] 2.3 Ship carrying capacity The daily average departure capacity of ships is based on the formula Calculations show that three core air routes will be opened in this scenario. , ships, , ships, , ships, (During peak hours), the calculated number is 138.04 flights per day.

[0090] The total daily transport capacity of a single vessel is based on the formula. Calculation, parameter values: Standard vehicle / shift (Ship-Dock Dynamic Coupling Coefficient) (Peak period ship operation efficiency index). (Results from NSGA-Ⅲ algorithm) The calculated capacity is 7218.05 TEUs / day, which is 48.8% of the port's resource capacity. Therefore, the shipping capacity is determined to be the core bottleneck of the current transportation system.

[0091] 2.4 Service Capacity of Rear Gates According to the formula Calculation, parameter values: , Vehicles, , Vehicles, , Vehicles, (Pure peak hours) (Pure peak scenario) Vehicles, (Peak-hour gate operation efficiency index).

[0092] The calculated capacity is 156,568 vehicles per day, indicating that the gate's service capacity far exceeds actual demand, with no capacity bottleneck.

[0093] 2.5 Capabilities in other aspects The calculated capacity of the inland collection and distribution channel facilities is 2,150 standard vehicles per hour, with an average daily throughput of 51,600 standard vehicles, which meets the actual daily throughput demand of 43,000 vehicles. The turnover capacity of the waiting area is 18,000 vehicles per day, which matches the unloading capacity of the wharf. The capacity of the wharf and transportation on the opposite bank is 13,200 standard vehicles per day, which is slightly lower than the capacity of the wharf on this side, and is a secondary weak link.

[0094] 3. Peak-period multi-level bottleneck identification and dynamic early warning Based on the facility capacity assessment results at each stage and real-time monitoring data throughout the entire process, a hybrid root cause model combining "causal inference - multi-level Bayesian network - fault tree analysis (FTA)" is used to locate bottlenecks at multiple levels during peak periods and trigger corresponding dynamic early warnings. 3.1 Multi-level causal root cause localization Indicator layer: Three core abnormal indicators were identified, namely, the ship capacity utilization rate of 98.5% (close to saturation), the average daily congestion time of the port's rear collection and distribution channel of 1.5 hours (exceeding the preset threshold of 1 hour), and the average waiting time of the waiting area of ​​45 minutes (exceeding the preset threshold of 30 minutes). At the link level, the correlation and tracing revealed three types of link shortcomings: insufficient ship carrying capacity (core shortcoming), low efficiency of port rear collection and distribution channel scheduling (secondary shortcoming), and unreasonable vehicle classification scheduling at the waiting yard (derived shortcoming). Causal Root Cause Layer: By eliminating spurious correlations through causal inference algorithms and combining FTA analysis to determine the root cause propagation path, the contribution of root causes is quantified: ① The core reason for insufficient shipping capacity is the insufficient number of vessels operating during peak periods (contribution 85%), and the secondary reason is the slight increase in route travel time due to sea conditions (contribution 15%); ② The direct cause of congestion in the port's rear access channels is mixed vehicle scheduling (contribution 70%), and the indirect cause is the underutilization of emergency lanes (contribution 30%); ③ Excessive waiting time at the ferry terminal is a derivative result of congestion in the port's rear access channels, contributing 100%. Impact on Evolution Layer: Predicting the evolution trend of the bottleneck. If the shipping capacity is not replenished in time, there will be a backlog of ship departure schedules in the middle of the Spring Festival travel rush, leading to saturation of the waiting area capacity, which in turn will exacerbate the congestion of the port's rear collection and distribution channels. The system resilience index will decrease at a rate of 5% / h. If only the collection and distribution channel scheduling is optimized, the waiting time can be reduced, but the core capacity gap problem cannot be solved.

[0095] 3.2 Dynamic Early Warning Triggering and Intervention Based on abnormal indicators and the evolution trend of shortcomings, combined with the "red-yellow-blue + trend" dual-dimensional early warning threshold, a yellow warning is triggered (the ship carrying capacity is close to 70% of the peak demand and is showing a continuous downward trend, and the congestion time of the port's rear collection and distribution channels exceeds the threshold of 50%).

[0096] The system automatically pushes early warning information to port and shipping management departments and transportation companies, identifies abnormal indicators, bottlenecks in the process, root causes and potential impacts, and initiates moderate automated intervention measures: ① Temporarily deploy two standby vessels to the core route to supplement shipping capacity; ② Open one emergency lane in the port's rear access road and implement a passenger and freight separation scheduling strategy; ③ Optimize vehicle type classification scheduling at the waiting area, set up a dedicated waiting area for dangerous goods vehicles, and improve turnover efficiency.

[0097] 3.3 Prioritization of Improvements for Shortcomings During Peak Periods A four-dimensional cost-effectiveness evaluation model based on "life cycle cost, environmental impact, social benefits, and operational benefits" was introduced to calculate the cost-effectiveness index of each improvement plan for shortcomings. The results are as follows: Supplementing shipping capacity: The improved resilience index increased by 35%, which can reduce the number of passengers stranded during peak periods by about 2,000 people per day, improve operational efficiency by 28%, have moderate total life cycle costs, and have a small incremental environmental impact, making it a first-level priority. Optimize the scheduling of port-related collection and distribution channels: The improved resilience index increases by 20%, alleviating traffic congestion, resulting in significant social benefits, no additional construction costs, and a negative environmental impact (reduced idling emissions). Its priority is Level 1. Expanding the waiting area for ferry crossings: Although it can improve turnover capacity, the total life cycle cost is high, and it is a passive improvement of derivative shortcomings, so its priority is level three; Expansion of the wharf on the opposite bank: This is a secondary weak link. Improvements can increase transportation capacity by 10%, but the construction period is long, so it is a secondary priority.

[0098] The system uses machine learning algorithms to associate historical Spring Festival travel rush cases, matching customized improvement paths for first-priority solutions, and forming the "Implementation Plan for Improving Shortcomings in Cross-Strait Passenger and Roll-on / Roll-off Transportation during the Spring Festival Travel Rush".

[0099] 4. Peak-period seven-dimensional dynamic resilience assessment The results of identifying weaknesses and tracing root causes are substituted into a seven-dimensional dynamic resilience assessment model, employing a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" (pure peak period weight: After standardizing each resilience index, the comprehensive resilience index for peak periods is calculated, and the resilience status assessment for each dimension is completed. 4.1 Calculation of resilience indices for each dimension (standardized values, ranging from [0,1]) Dynamic resilience: Peak resilience margin is -12% (capacity gap), time-period dynamic adaptation coefficient is 0.23, standardized value ; Disturbance resilience: Disturbance resilience coefficient 82% (resilience decreases by 18% under heavy fog disturbance), short-board disturbance transmission coefficient 1.8, normalized value. ; Resilience: Natural recovery time without emergency dispatch is 8 hours; recovery time after intervention is 3 hours; recovery efficiency coefficient is 20.8%; emergency dispatch contribution to recovery is 62.5%; standardized value. ; Resilience Reserves: Vessel capacity reserve coefficient 5% (close to saturation), emergency resource response efficiency 90%, standardized value. ; Collaborative resilience: The collaborative response completion time is 60% of the average independent response time, the collaborative resilience improvement factor is 25%, and the standardized value is... ; Intelligent early warning resilience: Early warning duration is 4 hours, and the resilience index increases by 18% after the early warning. (Standardized value) ; Resilience evolution trend: Before the warning, the resilience evolution rate was -5% / h; after the warning, the intervention implementation rate was +3% / h; the resilience stability coefficient was 0.2; and the standardized value was... .

[0100] 4.2 Calculation of Comprehensive Resilience Index The overall resilience index is 0.519, which is at a medium level. Dynamic resilience and resilience reserve are the main dimensions that drag down the index, while recovery and synergistic resilience are good. The intelligent early warning resilience has achieved the expected effect, which verifies the effectiveness of the early warning mechanism of this invention.

[0101] 5. Intelligent agent-driven end-to-end autonomous iterative closed-loop optimization Based on peak-period bottleneck improvement priorities, seven-dimensional resilience assessment results, and causal root causes, the intelligent agent autonomously generates and executes optimization solutions, completing a closed-loop autonomous iteration across the entire chain. The core process execution results are as follows: The intelligent agent generates two core optimization solutions for the primary priority bottlenecks: Solution 1 is "resettlement of backup vessels + optimization of route departure intervals" (supplementation of vessel capacity), and Solution 2 is "separation of passenger and freight channels in the collection and distribution channels + activation of emergency lanes" (improvement of dispatching efficiency). The feasibility of the solutions is verified through simulation models, and it is predicted that the comprehensive resilience index can be improved to above 0.75 after the implementation of the solutions. The optimization plan is executed automatically: The system is seamlessly integrated with the ship operation management system, port collection and distribution management system, and traffic control system, and automatically issues the optimization plan: ① The ship operation system receives the order to dispatch backup ships and completes the route deployment of 2 ships within 2 hours, shortening the departure interval of the core route from 20 minutes to 15 minutes; ② The collection and distribution management system and the traffic control system work together to open emergency lanes, implement passenger and freight separation, and update traffic signs in sync. Multi-dimensional real-time feedback: After the implementation of the solution, the following data was monitored and fed back in real time through the full-process data collection system: ① The ship carrying capacity increased to 9,850 TEUs / day, and the capacity gap decreased from 28% to 8%; ② The congestion time of the port's rear collection and distribution channels decreased from 1.5 hours to 0.5 hours, and the traffic speed increased to 40 km / h; ③ The average waiting time at the waiting area decreased from 45 minutes to 20 minutes, and the user waiting satisfaction increased to 92%; ④ The system's comprehensive resilience index increased from 0.519 to 0.786, reaching a good resilience level; ⑤ The unit transportation energy consumption decreased by 0.08 kWh / vehicle, and carbon emissions decreased by 12%, achieving a dual improvement in efficiency and green low-carbon development; Intelligent agent autonomous iterative optimization: Based on feedback data, the intelligent agent automatically adjusts the evaluation model parameters: ① Adjusting the ship scenario adjustment factor The value for the standby vessel deployment scenario is optimized to 0.15; ② The vehicle type mixing correction coefficient for the port rear collection and distribution channel is updated. The value is 0.88 in the passenger and freight separation scenario; ③ Optimize the weight of dynamic resilience in the seven-dimensional resilience assessment model to 0.3 (to adapt to the scenario after capacity supplementation), form an iterative assessment model and optimization strategy, and store it in the solution knowledge base; Results Output and Visualization: The intelligent interactive visualization decision support platform automatically generates the "Intelligent Monitoring and Resilience Assessment Report on Cross-Strait Passenger and Roll-on / Roll-off Transportation during the Spring Festival Travel Rush." ​​It uses map visualization to show the effects of congestion relief on collection and distribution channels, adjustments to ship route deployments, and optimization of vehicle distribution at waiting areas. It uses radar charts and trend curves to show changes in facility capacity at each stage, the trend of resilience index improvement, and the effects of improvement on shortcomings. It uses causal root cause diagrams to show the effects of intervention measures on eliminating root causes, providing managers with intuitive and comprehensive decision support.

[0102] 6. Implementation effect verification The implementation of the method of this invention in the cross-strait passenger and vehicle ferry transportation scenario during the Spring Festival travel rush has achieved the following core effects, verifying the scientific validity, accuracy, and practical guidance of the method: Monitoring and data fusion results: 100% full-process monitoring coverage, 60% improvement in data processing efficiency during peak periods, 98.6% accuracy in data completion, and ≤1 minute response time for intelligent linkage of five-dimensional data, meeting the dynamic monitoring needs during peak periods; Facility capacity assessment results: The quantitative error of facility capacity during peak periods in each stage is ≤5%, accurately identifying the core bottleneck of shipping capacity and providing a quantitative basis for resource allocation; Shortcomings identification and early warning effects: Multi-level causal root cause identification accuracy is 95%, dynamic early warning trigger lead time is ≥4h, and automated intervention response time is ≤2h, effectively avoiding the resilience decay of the transportation system during peak periods; Resilience assessment results: The seven-dimensional dynamic resilience assessment comprehensively quantifies the system's resilience status, and the overall resilience index matches the actual operating status with a 90% degree of accuracy, providing scientific guidance for resilience improvement; Closed-loop optimization results: After the implementation of the optimization plan, the core shortcomings were effectively addressed, the overall system resilience index increased by 51.4%, peak-hour transportation efficiency increased by 28%, user satisfaction increased by 12.2%, and unit transportation energy consumption decreased by 13.3%, achieving multi-objective optimization of "efficiency-safety-greenness-resilience".

[0103] This embodiment demonstrates that the intelligent monitoring and dynamic resilience assessment method for the entire process of cross-strait passenger and vehicle ferry transportation of the present invention can effectively adapt to the differentiated needs of peak-season transportation scenarios, solve the core pain points of existing technologies, and provide comprehensive scientific support for peak-season operation management, emergency dispatch, risk prevention and control, and resilience enhancement of cross-strait passenger and vehicle ferry transportation. It has good engineering practicality and promotion value.

[0104] Example 3 Based on the same inventive concept, the present invention also provides an intelligent monitoring and dynamic resilience assessment system for the entire process of cross-strait passenger and roll-on / roll-off transportation, used to implement the methods described in the foregoing embodiments. The system includes: a data acquisition and processing module, a capacity assessment module, a shortcoming location and root cause tracing module, a resilience assessment module, and a closed-loop optimization module. The data acquisition and processing module is used to collect multi-source data of the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to build a five-dimensional dynamic relational database. The capacity assessment module is used to obtain the peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm. The shortcoming location and root cause tracing module is used to obtain the shortcoming location and root cause tracing results based on the full-process peak period facility capacity assessment results and the five-dimensional dynamic correlation database, through a hybrid root cause model. The resilience assessment module is used to obtain resilience assessment results based on the results of shortcoming location and root cause tracing and the seven-dimensional dynamic resilience assessment model, using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment". The closed-loop optimization module is used to construct a fully autonomous iterative closed-loop process driven by intelligent agents. Based on the resilience assessment results, it enables the autonomous generation, execution, feedback, and iteration of optimization schemes during peak periods.

[0105] Furthermore, in this embodiment, the process of intelligent preprocessing of the collected multi-source data includes: Edge computing technology is used to preprocess the collected multi-source data in real time, filter key data fields, remove invalid data and abnormal fluctuation values, and obtain filtered data. A data quality assessment model is constructed based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". The data quality assessment model is used to clean the screened data to obtain cleaned data. The cleaned data is repaired by using the methods of "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction" to obtain complete data; A four-dimensional dynamic weight model is constructed based on "data source credibility, data timeliness, scenario importance, and causal relationship strength" to obtain the weight values ​​of each supplementary data.

[0106] Furthermore, in this embodiment, the method for obtaining the peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, includes: Port back-end collection and distribution channel capacity: ; in, : Peak-period capacity of port rear access and distribution channels; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor; : Mixed vehicle model correction factor; Average headway of vehicles; Average parking distance between vehicles; Channel scenario adjustment factor; Peak traffic flow correction factor; Route resource capabilities: ; in, Peak season route resource capacity; Number of shipping routes opened at the port; Maximum number of vehicles / passengers that a single ship can carry; Shortest departure interval during peak hours; : Coupling optimization function; Carbon footprint throughout its entire life cycle; Port resource capacity: Average daily departure capacity per berth: ; Total daily transport capacity of a single-sided wharf: ; in, Average loading and unloading time during peak hours at a single-sided wharf; Average berthing and departure time at peak times for a single-sided pier; Average interval between adjacent vessels during peak periods; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles a single vessel can carry; : Adjustment factor for berth scenarios; : Dock-ship dynamic coupling coefficient; Peak-hour terminal operation efficiency index; : Terminal-operation coupling optimization function; Shipping capacity: Average daily departure capacity of vessels: ; Total daily vehicle transport capacity of a single vessel: ; in, : No. Total transit time for a single vessel on this route; : No. Number of vessels operating on each route during peak periods; Number of routes; Ship scenario adjustment factor; : Ship-dock dynamic coupling coefficient; Peak-period ship operation efficiency index; Ship-operation coupling optimization function; Supporting facilities - service capacity of the rear gate: ; in, , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These represent the average time for trucks, passenger vehicles, and hazardous materials transport vehicles to clear customs during peak hours. : Gate scene adjustment factor; Energy consumption per unit of transport at the gate; Energy consumption weighting factor; Peak-hour gate operation efficiency index.

[0107] Furthermore, in this embodiment, the hybrid root cause model includes: an indicator layer, a link layer, a causal root cause layer, and an influence evolution layer; The indicator layer is used to identify abnormal indicators by comparing real-time monitoring data during peak periods with preset thresholds. The aforementioned process layer is used to trace the corresponding process weaknesses based on the correlation analysis of abnormal indicators. The causal root cause layer is used to identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithm and eliminate spurious correlations; then, the propagation path of the root cause is analyzed through FTA and the contribution of the root cause is quantified by combining Bayesian network. The impact evolution layer is used to combine the characteristics of peak transportation scenarios to predict the degree of impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios.

[0108] Furthermore, in this embodiment, based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, the process of obtaining the resilience assessment results using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" includes: Dynamic toughness assessment: Peak resilience margin: ; Time-based dynamic adaptation coefficient: ; in, The overall carrying capacity of the system during peak hours; Actual transportation demand during peak hours; , These represent the overall system capacity and actual demand for each hour during peak periods; Disturbance resilience assessment: Disturbance immunity coefficient: ; Short-board disturbance transmission coefficient: ; in, , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. Capacity loss in bottleneck links during peak periods; Total system capacity loss during peak periods; Recovery assessment: Recovery efficiency coefficient: ; Contribution to emergency dispatch and recovery: ; in, The time it takes for the system to recover to normal capacity after a peak-period disturbance; Natural recovery time during peak periods without emergency dispatch; Resilience reserve capacity assessment: Capacity / Road Reserve Coefficient: ; Emergency resource response efficiency: ; in, Maximum carrying capacity of core facilities; The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; Synergistic resilience assessment: Cooperative response efficiency coefficient: ; Collaborative resilience enhancement coefficient: ; Intelligent early warning resilience assessment: ; Resilience evolution trend assessment: Resilience evolution rate: ; Toughness stability coefficient: ; in, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period; Peak-period comprehensive resilience index: ; in, , , , , , , : Assign optimal weights to the seven resilience dimensions respectively. , , , , , , : These are the standardized values ​​of each resilience index.

[0109] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for intelligent monitoring and dynamic resilience assessment of the entire process of cross-strait passenger and roll-on / roll-off (Ro-Ro) transportation, characterized in that, The method includes: Collect multi-source data from the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to construct a five-dimensional dynamic relational database; Based on a five-dimensional dynamic relational database and a facility capacity coupled optimization evaluation model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, the peak-period facility capacity evaluation results are obtained. Based on the full-process peak-period facility capacity assessment results and the five-dimensional dynamic correlation database, the results of shortcoming location and root cause tracing are obtained through a hybrid root cause model. Based on the results of shortcoming identification and root cause tracing and the seven-dimensional dynamic resilience assessment model, a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to obtain the resilience assessment results. Construct a closed-loop process for intelligent agent-driven end-to-end autonomous iteration, and based on resilience assessment results, realize the autonomous generation, execution, feedback and iteration of optimization solutions during peak periods; Methods for obtaining peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, include: Port back-end collection and distribution channel capacity: ； in, : Peak-period capacity of port rear access and distribution channels; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor; : Mixed vehicle model correction factor; Average headway of vehicles; Average parking distance between vehicles; Channel scenario adjustment factor; Peak traffic flow correction factor; Route resource capabilities: ； in, Peak season route resource capacity; Number of shipping routes opened by the port; Maximum number of vehicles / passengers that a single ship can carry; Shortest departure interval during peak hours; : Coupling optimization function; Carbon footprint throughout its entire life cycle; Port resource capacity: Average daily departure capacity per berth: ； Total daily transport capacity of a single-sided wharf: ； in, Average loading and unloading time during peak hours at a single-sided wharf; Average berthing and departure time at peak times for a single-sided pier; Average interval between adjacent vessels during peak periods; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles a single vessel can carry; : Adjustment factor for berth scenarios; : Dock-ship dynamic coupling coefficient; Peak-hour terminal operation efficiency index; : Terminal-operation coupling optimization function; Shipping capacity: Average daily departure capacity of vessels: ； Total daily vehicle transport capacity of a single vessel: ； in, : No. Total transit time for a single vessel on this route; : No. Number of vessels operating on each route during peak periods; Number of routes; Ship scenario adjustment factor; : Ship-dock dynamic coupling coefficient; Peak-period ship operation efficiency index; Ship-operation coupling optimization function; Supporting facilities - service capacity of the rear gate: ； in, , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These represent the average time for trucks, passenger vehicles, and hazardous materials transport vehicles to clear customs during peak hours. : Gate scene adjustment factor; Energy consumption per unit of transport at the gate; Energy consumption weighting factor; Peak-hour gate operation efficiency index.

2. The method according to claim 1, characterized in that, Methods for intelligent preprocessing of collected multi-source data include: Edge computing technology is used to preprocess the collected multi-source data in real time, filter key data fields, remove invalid data and abnormal fluctuation values, and obtain filtered data. A data quality assessment model is constructed based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". The data quality assessment model is used to clean the screened data to obtain cleaned data. The cleaned data is repaired by using the methods of "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction" to obtain complete data; A four-dimensional dynamic weight model is constructed based on "data source credibility, data timeliness, scenario importance, and causal relationship strength" to obtain the weight values ​​of each supplementary data.

3. The method according to claim 1, characterized in that, The hybrid root cause model includes: indicator layer, link layer, causal root cause layer, and influence evolution layer; The indicator layer is used to identify abnormal indicators by comparing real-time monitoring data during peak periods with preset thresholds. The aforementioned process layer is used to trace the corresponding process weaknesses based on the correlation analysis of abnormal indicators. The causal root cause layer is used to identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithm and eliminate spurious correlations; then, the propagation path of the root cause is analyzed through FTA and the contribution of the root cause is quantified by combining Bayesian network. The impact evolution layer is used to combine the characteristics of peak transportation scenarios to predict the degree of impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios.

4. The method according to claim 1, characterized in that, Based on the results of shortcoming identification and root cause tracing, and a seven-dimensional dynamic resilience assessment model, a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" is adopted to obtain resilience assessment results. The methods include: Dynamic toughness assessment: Peak resilience margin: ； Time-based dynamic adaptation coefficient: ； in, The overall carrying capacity of the system during peak hours; Actual transportation demand during peak hours; , These represent the overall system capacity and actual demand for each hour during peak periods; Disturbance resilience assessment: Disturbance immunity coefficient: ； Short-board disturbance transmission coefficient: ； in, , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. Capacity loss in bottleneck links during peak periods; Total system capacity loss during peak periods; Recovery assessment: Recovery efficiency coefficient: ； Contribution to emergency dispatch and recovery: ； in, The time it takes for the system to recover to normal capacity after a peak-period disturbance; Natural recovery time during peak periods without emergency dispatch; Resilience reserve capacity assessment: Capacity / Road Reserve Coefficient: ； Emergency resource response efficiency: ； in, Maximum carrying capacity of core facilities; The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; Synergistic resilience assessment: Cooperative response efficiency coefficient: ； Collaborative resilience enhancement coefficient: ； Intelligent early warning resilience assessment: ； Resilience evolution trend assessment: Resilience evolution rate: ； Toughness stability coefficient: ； in, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period; Peak-period comprehensive resilience index: ； in, , , , , , , : Assign optimal weights to the seven resilience dimensions respectively. , , , , , , : These are the standardized values ​​of each resilience index.

5. A smart monitoring and dynamic resilience assessment system for the entire process of cross-strait passenger and roll-on / roll-off transportation, the system being used to implement the method described in any one of claims 1-4, characterized in that, The system includes: a data acquisition and processing module, a capability assessment module, a weakness identification and root cause tracing module, a resilience assessment module, and a closed-loop optimization module; The data acquisition and processing module is used to collect multi-source data of the entire process of cross-strait passenger and roll-on / roll-off transportation and perform intelligent preprocessing to build a five-dimensional dynamic relational database. The capacity assessment module is used to obtain the peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm. The shortcoming location and root cause tracing module is used to obtain the shortcoming location and root cause tracing results based on the full-process peak period facility capacity assessment results and the five-dimensional dynamic correlation database, through a hybrid root cause model. The resilience assessment module is used to obtain resilience assessment results based on the results of shortcoming location and root cause tracing and the seven-dimensional dynamic resilience assessment model, using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment". The closed-loop optimization module is used to construct a fully autonomous iterative closed-loop process driven by intelligent agents. Based on the resilience assessment results, it enables the autonomous generation, execution, feedback, and iteration of optimization schemes during peak periods.

6. The system according to claim 5, characterized in that, The process of intelligent preprocessing of collected multi-source data includes: Edge computing technology is used to preprocess the collected multi-source data in real time, filter key data fields, remove invalid data and abnormal fluctuation values, and obtain filtered data. A data quality assessment model is constructed based on a hybrid model of "attention mechanism + Transformer + spatiotemporal feature transfer". The data quality assessment model is used to clean the screened data to obtain cleaned data. The cleaned data is repaired by using the methods of "historical similar scene data completion + real-time neighborhood data calibration + cross-dimensional data verification + causal relationship correction" to obtain complete data; A four-dimensional dynamic weight model is constructed based on "data source credibility, data timeliness, scenario importance, and causal relationship strength" to obtain the weight values ​​of each supplementary data.

7. The system according to claim 5, characterized in that, Methods for obtaining peak-period facility capacity assessment results based on a five-dimensional dynamic relational database and a facility capacity coupled optimization assessment model, combined with the NSGA-Ⅲ algorithm and the DQN algorithm, include: Port back-end collection and distribution channel capacity: ； in, : Peak-period capacity of port rear access and distribution channels; : Number of effective lanes in the passage; Lane width correction factor; Road condition correction factor; : Mixed vehicle model correction factor; Average headway of vehicles; Average parking distance between vehicles; Channel scenario adjustment factor; Peak traffic flow correction factor; Route resource capabilities: ； in, Peak season route resource capacity; Number of shipping routes opened by the port; Maximum number of vehicles / passengers that a single ship can carry; Shortest departure interval during peak hours; : Coupling optimization function; Carbon footprint throughout its entire life cycle; Port resource capacity: Average daily departure capacity per berth: ； Total daily transport capacity of a single-sided wharf: ； in, Average loading and unloading time during peak hours at a single-sided wharf; Average berthing and departure time at peak times for a single-sided pier; Average interval between adjacent vessels during peak periods; Number of passenger and roll-on / roll-off berths in operation; Maximum number of vehicles a single vessel can carry; : Adjustment factor for berth scenarios; : Dock-ship dynamic coupling coefficient; Peak-hour terminal operation efficiency index; : Terminal-operation coupling optimization function; Shipping capacity: Average daily departure capacity of vessels: ； Total daily vehicle transport capacity of a single vessel: ； in, : No. Total transit time for a single vessel on this route; : No. Number of vessels operating on each route during peak periods; Number of routes; Ship scenario adjustment factor; : Ship-dock dynamic coupling coefficient; Peak-period ship operation efficiency index; Ship-operation coupling optimization function; Supporting facilities - service capacity of the rear gate: ； in, , , These refer to the number of dedicated gates for freight cars, passenger cars, and hazardous materials transport vehicles during peak hours. , , These represent the average time for trucks, passenger vehicles, and hazardous materials transport vehicles to clear customs during peak hours. : Gate scene adjustment factor; Energy consumption per unit of transport at the gate; Energy consumption weighting factor; Peak-hour gate operation efficiency index.

8. The system according to claim 5, characterized in that, The hybrid root cause model includes: an indicator layer, a link layer, a causal root cause layer, and an influence evolution layer; The indicator layer is used to identify abnormal indicators by comparing real-time monitoring data during peak periods with preset thresholds. The aforementioned process layer is used to trace the corresponding process weaknesses based on the correlation analysis of abnormal indicators. The causal root cause layer is used to identify the causal relationship between "indicator anomaly - link weakness - root cause" through causal inference algorithm and eliminate spurious correlations; then, the propagation path of the root cause is analyzed through FTA and the contribution of the root cause is quantified by combining Bayesian network. The impact evolution layer is used to combine the characteristics of peak transportation scenarios to predict the degree of impact, diffusion path, and resilience evolution trend of bottlenecks on different future scenarios.

9. The system according to claim 5, characterized in that, Based on the results of shortcoming identification and root cause tracing, and a seven-dimensional dynamic resilience assessment model, the process of obtaining resilience assessment results using a dynamic weight allocation mechanism of "Bayesian optimization + scenario risk clustering + causal weight adjustment" includes: Dynamic toughness assessment: Peak resilience margin: ； Time-based dynamic adaptation coefficient: ； in, The overall carrying capacity of the system during peak hours; Actual transportation demand during peak hours; , These represent the overall system capacity and actual demand for each hour during peak periods; Disturbance resilience assessment: Disturbance immunity coefficient: ； Short-board disturbance transmission coefficient: ； in, , These represent the system's comprehensive carrying capacity under normal operating conditions during peak periods and under disturbance scenarios during peak periods, respectively. Capacity loss in bottleneck links during peak periods; Total system capacity loss during peak periods; Recovery assessment: Recovery efficiency coefficient: ； Contribution to emergency dispatch and recovery: ； in, The time it takes for the system to recover to normal capacity after a peak-period disturbance; Natural recovery time during peak periods without emergency dispatch; Resilience reserve capacity assessment: Capacity / Road Reserve Coefficient: ； Emergency resource response efficiency: ； in, Maximum carrying capacity of core facilities; The actual carrying capacity of core facilities during peak periods; Standard time for emergency resource dispatch during peak periods; : Actual dispatch time of emergency resources during peak periods; Synergistic resilience assessment: Cooperative response efficiency coefficient: ； Collaborative resilience enhancement coefficient: ； Intelligent early warning resilience assessment: ； Resilience evolution trend assessment: Resilience evolution rate: ； Toughness stability coefficient: ； in, , : These are the system resilience indices at peak times t1 and t2, respectively; , , : These represent the maximum, minimum, and average values ​​of the system resilience index during the peak monitoring period; Peak-period comprehensive resilience index: ； in, , , , , , , : Assign optimal weights to the seven resilience dimensions respectively. , , , , , , : These are the standardized values ​​of each resilience index.

Images